摘要
为解决当前图像匹配算法忽略了相邻图像层次间灰度量级的差异性导致的较多的误检与漏检现象,使其匹配精度不高的问题,本文设计了基于差分制约模型与三角网优化的图像匹配技术.首先,利用差分高斯函数来构造差分制约方法,对相邻图像层次间的灰度量级进行一致性约束,准确提取图像特征点;然后,通过计算圆形邻域内的Haar小波响应值,确定特征点的主方向;再计算圆形邻域内的梯度与灰度特征,得到相应的特征向量;利用主方向与特征向量来生成实现特征点描述符.利用特征点描述符求取特征点之间欧氏距离的最近邻与次近邻比值,对图像特征完成初步匹配;最后,通过初匹配之间的空间关系构建三角网约束规则,对错误匹配特征点进行剔除,对匹配结果进行优化.实验结果表明:与当前图像匹配算法相比,所提算法具有更高的匹配正确度与鲁棒性.
At present many image matching algorithm mainly through to image feature detection and image matching in each image layer by using Hessian matrix, because the method ignores the differences between adjacent levels of image gray level of the image feature detection, missing a lot of feature points and there are more error detection. An image matching algorithm based on differential constraint model coupling triangulation constraint was proposed in this paper. First, the differential Gauss function is used to construct the differential control model, which is used to congrate the gray level between adjacent images, so as to achieve fast and accurate extraction of image feature points. Then, we calculate the main direction of the feature points by calculating the Haar wavelet response value in the circular neighborhood, then we get the gradient and grayscale feature in the circular neighborhood to get the feature vectors, so as to achieve the generation of the feature point descriptors. Finally, using the feature point descriptor for the minimum Euclidean distance between feature points and small Euclidean distance to match feature points, then using the matching space relationship between the feature points of constructing triangulation constraint rules, to remove the false feature points matching, thus completing the image matching. The simulation results and analysis show that: compared with the current image matching algorithm, the algorithm designed in this paper has higher matching accuracy and matching efficiency.
作者
黄源
张福泉
HUANG Yuan;ZHANG Fuquan(Department of Computer,Chongqing Aerospace Career Technical College,Chongqing,400021,China;School of Software,Beijing Institute of Technology,Beijing,100081,China)
出处
《新疆大学学报（自然科学版）》
CAS
2018年第4期437-444,共8页
Journal of Xinjiang University(Natural Science Edition)
基金
重庆市教育委员会科学技术研究计划青年项目资助(KJQN201803002)
关键词
图像匹配
差分制约模型
HAAR小波
欧氏距离
差分高斯函数
三角网约束规则
image matching
differential constraint model
Haar wavelet
Euclidean distance
difference of Gaussian
triangulation constraint rules
